Welcome to Locator’s documentation!

Locator is a deep learning-based tool for predicting geographic coordinates from genotype matrices. It uses TensorFlow and Keras to build models that can accurately predict the geographic origin of samples based on their genetic data.

Contents:

• Installation
  • Requirements
  • Installing with pixi (recommended)
  • Installing with pip
  • Installing from Source
  • Verifying Installation
  • Next Steps
• Usage Guide
  • Basic Usage
  • Training and Prediction
  • Holdout Analysis
  • Ensemble Models
  • Windowed Analysis
  • Jacknife Analysis
  • Using Range Masks
  • Memory-Efficient Data Pipeline
  • GPU Configuration
  • Data Augmentation
  • Handling Missing Coordinates
  • Multi-GPU Parallel Analysis
  • Next Steps
• CLI Guide
  • Basic Usage
  • Uncertainty and Windowed Analysis
  • Bootstraps
  • Jacknife
  • See Also
• Ensemble Models Guide
  • Basic Ensemble Training
  • Making Predictions
  • Advanced Features
  • Model Persistence
  • See Also
• The PCA Model
  • When to use it
  • Basic usage
  • How it works
  • Choosing how many components to keep
  • Settings
  • When you cannot use it
• Parallel Analysis Guide
  • When to Use Parallel Analysis
  • Installation
  • Quick Start
  • Analysis Functions
  • GPU Fraction Settings
  • Parallel Ensemble Training
  • See Also
• Plotting Guide
  • Prediction Visualization
  • Error Analysis
  • Sample Weight Visualization
  • Jupyter Notebook Integration
  • See Also
• Handling Missing Coordinates Guide
  • Understanding NA Samples
  • NA Action Modes
  • Checking Your Data
  • Method-Specific Behavior
• API Reference
  • Core Module
  • Ensemble Functionality
  • Models Module
  • Data Module
  • Sample Weights Module
  • GPU Optimizer Module
  • Internal Modules (Implementation Details)
  • Parallel Analysis Module
  • Plotting Module
  • Configuration Options
• Contributing
  • Development Setup
  • Code Style
  • Testing
  • Making Changes
  • Commit Messages
  • Documentation
  • Reporting Issues
  • License

Quick Links

• Installation - Installation instructions

• Usage Guide - Basic and advanced usage guide

• CLI Guide - Command-line interface

• Ensemble Models Guide - Ensemble models and k-fold cross-validation

• The PCA Model - The PCA model for data with many SNPs

• Parallel Analysis Guide - Multi-GPU parallel analysis guide

• Plotting Guide - Visualization and plotting guide

• Handling Missing Coordinates Guide - Guide for handling missing coordinates

• API Reference - Complete API reference

• Contributing - Contribution guidelines

Key Features

• Deep learning-based coordinate prediction

• GPU acceleration with automatic optimization

• Mixed precision training for 2x speedup

• Memory-efficient data pipeline with zero-copy splits

• Multi-GPU parallel analysis with Ray

• Species range mask integration

• Consistent handling of missing coordinates

• Built-in data augmentation

Getting Started

1. Installation - Install Locator

2. Usage Guide - Learn the basics

3. API Reference - Explore the API

For Developers

• Contributing - How to contribute

• API Reference - Detailed API documentation

Indices and tables

• Index

• Module Index

• Search Page